http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48049

           Summary: 4.6 Regression? DW_AT_accessibility change breaks GDB
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: minor
          Priority: P3
         Component: debug
        AssignedTo: unassig...@gcc.gnu.org
        ReportedBy: jan.kratoch...@redhat.com
                CC: ja...@gcc.gnu.org
            Target: x86_64-fedora15-linux-gnu


Only -gdwarf-3 is affected here:

debug/45124 has changed DW_AT_accessibility generation for DWARF-3+.

GGC-4.5 was compatible with GDB but both were not compliant with DWARF-3+.
GCC-4.6 is now DWARF-3+ compliant but it breaks GDB.

By providing DW_AT_accessibility explicitly even for cases when one could
assume a default value may be both compatible with existing GDBs and DWARF-3+
compliant.
I believe this was the Jakub's intention but it did not get implemented in GCC
that way:
Bug 45124 Comment 1:
> I'm afraid we need to live with some extra DW_AT_accessibility attributes for 
> > compatibility.

class C { private: int private_; } v;
gcc -c -Wall -gdwarf-3
GNU gdb (GDB) 7.2.50.20110309-cvs

PASS: g++ (GCC) 4.5.3 20110124 (prerelease)
(gdb) ptype C
type = class C {
    int private_;
}
 <2><37>: Abbrev Number: 3 (DW_TAG_member)
    <38>   DW_AT_name        : (indirect string, offset: 0x2e): private_        
    <3c>   DW_AT_decl_file   : 1        
    <3d>   DW_AT_decl_line   : 4        
    <3e>   DW_AT_type        : <0x45>   
    <42>   DW_AT_data_member_location: 0        
    <43>   DW_AT_accessibility: 3       (private)

but:
FAIL: g++ (GCC) 4.6.0 20110309 (experimental)
(gdb) ptype C
type = class C {
  public:
  ^^^^^^^
    int private_;
}
because:
 <2><37>: Abbrev Number: 3 (DW_TAG_member)
    <38>   DW_AT_name        : (indirect string, offset: 0x54): private_        
    <3c>   DW_AT_decl_file   : 1        
    <3d>   DW_AT_decl_line   : 4        
    <3e>   DW_AT_type        : <0x44>   
    <42>   DW_AT_data_member_location: 0        


(That GDB should be changed is sure but there may be some DW_AT_producer
checks, depending on how it gets resolved for GCC-4.6.0.)

Reply via email to